Devices, systems, and methods for automated operation of infusion pumps

Abstract

A system for operation of an infusion pump may include a device configured to generate dosing functions for delivery of a medicament via an infusion pump. The system may also include the infusion pump, where the infusion pump is configured to obtain the dosing functions from the aforenoted device, determine state scores, determine a combined dosing function based on the dosing functions and the state scores, and deliver a dose of medicament to the user according to the combined dosing function. Each of the state scores represents a probability that a user is in a respective one of activity states. The user is in one of the activity states at any given time and is capable of transitioning among the activity states during a certain time period (e.g., in a given day).

Description

PATENT APPLICATION Devices, Systems, and Methods for Automated Operation of Infusion Pumps RELATED APPLICATIONS

[0001] The present application claims priority to U.S. Provisional App. No. 63 / 615,481(filed Dec.28, 2023), which is hereby incorporated in its entirety. TECHNICALFIELD

[0002] The present disclosure relates, generally, to infusion pumps and, more specifically,to automated operation of infusion pumps. BACKGROUND

[0003] There are a wide variety of medical treatments that include the administration of atherapeutic fluid in precise, known amounts at predetermined intervals. Devices and methods exist that are directed to the delivery of such fluids, which may be liquids or gases, are known in the art.

[0004] One category of such fluid delivery devices includes insulin injecting pumpsdeveloped for administering insulin to patients afflicted with type 1, or in some cases, type 2 diabetes. Some insulin injecting pumps are configured as portable or ambulatory infusion devices can provide continuous subcutaneous insulin injection and / or infusion therapy as an alternative to multiple daily injections of insulin via a syringe or an insulin pen. Such pumps are worn by the user and may use replaceable cartridges. In some embodiments, these pumps may also deliver medicaments other than, or in addition to, insulin, such as glucagon, pramlintide, and the like. Examples of such pumps and various features associated therewith include those disclosed in U.S. Patent App. Pub. Nos.2013 / 0324928 and 2013 / 0053816, U.S. Patent Nos.8,287,495, 8,573,027, 8,986,253, and 9,381,297, as well as PCT Patent App. No. PCT / US23 / 82084, each of which is incorporated herein by reference in its entirety.

[0005] Ambulatory infusion pumps for delivering insulin or other medicaments can be usedin conjunction with blood glucose monitoring systems, such as blood glucose meters (BGMs) and continuous glucose monitors (CGMs). A CGM provides a substantially continuous estimated blood glucose level through a transcutaneous sensor that estimates blood analyte levels, such as blood glucose levels, via the patient’s interstitial fluid CGM systems typically consist of a transcutaneously placed sensor, a transmitter, and a monitor.

[0006] Ambulatory infusion pumps typically allow the patient or caregiver to adjust theamount of insulin or other medicament delivered, by a basal rate or a bolus, based on bloodPATENT APPLICATION glucose data obtained by a BGM or a CGM, and in some cases include the capability to automatically adjust such medicament delivery. Some ambulatory infusion pumps may include the capability to interface with a BGM or CGM such as, e.g., by receiving measured or estimated blood glucose levels and automatically adjusting or prompting the user to adjust the level of medicament being administered or planned for administration or, in cases of abnormally low blood glucose readings, reducing or automatically temporarily ceasing or prompting the user temporarily to cease or reduce insulin administration. These portable pumps may incorporate a BGM or CGM within the hardware of the pump or may communicate with a dedicated BGM or CGM via wired or wireless data communication protocols, directly and / or via a device such as a smartphone. One example of integration of infusion pumps with CGM devices is described in U.S. Patent App. Pub. No.2014 / 0276419, which is hereby incorporated by reference herein.

[0007] As noted above, insulin or other medicament dosing by basal rate and / or bolustechniques could automatically be provided by a pump based on readings received into the pump from a CGM device that is, e.g., external to the portable insulin pump or integrated with the pump as a pump-CGM system in a closed-loop or semi-closed-loop fashion. With respect to insulin delivery, some systems including this feature can be referred to as artificial pancreas systems or dynamic artificial pancreas (DAP) system, because the systems serve to mimic biological functions of the pancreas for patients with diabetes. Such systems are also referred to as automated insulin delivery (AID) systems.

[0008] An AID system uses measurements of metabolic signals such as interstitial glucoseand user inputs such as carbohydrate entries to determine the optimal amount of insulin to deliver to maintain user blood glucose as close as possible to the euglycemic range. Current AID systems employ relatively simple algorithms that calculate insulin doses in a manner consistent with the limited capabilities of ambulatory infusion pumps. In addition, current systems have relied heavily on user inputs such as carb entries or exercise mode to appropriately treat large glycemic disturbances such as eating or exercise. SUMMARY

[0009] Embodiments of the present disclosure provide apparatuses and methods forautomated insulin delivery.

[0010] In some embodiments, a system for operation of an infusion pump includes a dosingfunction device and an infusion pump. The dosing function device is configured to generate a predictive model to simulate effects of dosing decisions over time. The dosing function devicePATENT APPLICATION is also configured to determine activity states based on the predictive model. A user is in one of the activity states at any given time and is capable of transitioning among the activity states during a certain time period. Additionally, the dosing function device is configured to determine dosing functions, each of the dosing functions representing dosing targets associated with a respective one of the activity states. The infusion pump is configured to obtain the dosing functions from the dosing function device, and determine state scores, each of the state scores representing a probability that the user is in a respective one of the activity states. The infusion pump is also configured to determine a combined dosing function based on the dosing functions and the state scores and deliver a dose of medicament to the user according to the combined dosing function.

[0011] In some embodiments, an infusion pump includes a processor and a non-transitory,computer-readable medium storing instructions which, when executed by the processor, cause the infusion pump to perform operations. The operations include receiving a set of dosing functions from a dosing function device, each dosing function representing dosing targets associated with a respective one of activity states. A user is in one of the activity states at any given time and is capable of transitioning among the activity states during a certain time period. The operations also include determining state scores, each of the state scores representing a probability that the user is in a respective one of the activity states. Further, the operations include determining a combined dosing function based on the dosing functions and the state scores, and delivering a dose of medicament to the user according to the combined dosing function.

[0012] In some embodiments, a computer-implemented method for operation of aninfusion pump includes receiving a set of dosing functions at the infusion pump and from a dosing function device, each dosing function representing dosing targets associated with a respective one of activity states. A user is in one of the activity states at any given time, and is capable of transitioning among the activity states during a certain time period. The computer- implemented method also includes determining state scores at the infusion pump, each of the state scores representing a probability that the user is in a respective one of the activity states. Further, the computer-implemented method includes determining, at the infusion pump, a combined dosing function based on the dosing functions and the state scores, and delivering a dose of medicament to the user via the infusion pump and according to the combined dosing function.

[0013] In some embodiments, a non-transitory, computer-readable medium storesinstructions which, when executed by a processor of an electronic device, cause the electronicPATENT APPLICATION device to perform operations. The operations include receiving a set of dosing functions from a dosing function device, each dosing function representing dosing targets associated with a respective one of activity states. A user is in one of the activity states at any given time, and is capable of transitioning among the activity states during a certain time period. The operations also include determining state scores, each of the state scores representing a probability that the user is in a respective one of the activity states. Further, the operations include determining a combined dosing function based on the dosing functions and the state scores, and delivering a dose of medicament to the user according to the combined dosing function.

[0014] The above summary is not intended to describe each illustrated embodiment orevery implementation of the subject matter hereof. The figures and the detailed description that follow more particularly exemplify various embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The invention may be more completely understood in consideration of the followingdetailed description of various embodiments of the invention in connection with the accompanying drawings, in which:

[0016] Figure 1 is a medical device that can be used with embodiments of the disclosure,according to various embodiments of the present disclosure.

[0017] Figure 2 is a block diagram representing a medical device that can be used withembodiments of the disclosure, according to various embodiments of the present disclosure.

[0018] Figures 3A-3B depict an embodiment of a pump system, according to variousembodiments of the present disclosure.

[0019] Figure 4 is a schematic representation of a system, according to variousembodiments of the present disclosure.

[0020] Figure 5 is a block diagram of a system for employing an automated glycemiccontrol algorithm, according to various embodiments of the present disclosure.

[0021] Figure 6 is a flow chart of a method for delivering optimal insulin doses, accordingto various embodiments of the present disclosure.

[0022] Figure 7 illustrates a hidden Markov model (HMM) used for estimating meal statesof a user, according to various embodiments of the present disclosure.

[0023] Figure 8 illustrates dependence of exemplary emission probability distributions onglucose influx of a user at different meal states, according to various embodiments of the present disclosure.PATENT APPLICATION

[0024] Figure 9A is a graph of a policy, according to various embodiments of the presentdisclosure.

[0025] Figure 9B is a graph of the policy of Figure 9A with a flat portion, a quick riseportion, and a slow rise portion identified, according to various embodiments of the present disclosure.

[0026] Figure 9C is a graph of the policy of Figure 9A with a max flat glucose and maxdose identified, according to various embodiments of the present disclosure.

[0027] Figure 9D is a graph of the policy of Figure 9A with indications of the impacts ofadjusting automated glycemic control algorithm parameters identified, according to various embodiments of the present disclosure.

[0028] Figure 10 is a graph showing different policies corresponding to different mealstates, according to various embodiments of the present disclosure.

[0029] Figure 11 illustrates optimal insulin dosing targets, according to variousembodiments of the present disclosure.

[0030] Figure 12 is a graph showing different dosing functions corresponding to differentmeal states, according to various embodiments of the present disclosure.

[0031] Figure 13 illustrates a process for blending different dosing functions at a giventime step, according to various embodiments of the present disclosure.

[0032] Figure 14 is a flow chart of a method for delivering a dose of medicament based ona combined dosing function, according to various embodiments of the present disclosure. DETAILED DESCRIPTION

[0033] The following detailed description should be read with reference to the drawings inwhich similar elements in different drawings are numbered the same. The drawings, which are not necessarily to scale, depict illustrative embodiments and are not intended to limit the scope of the invention.

[0034] One objective of the present teaching is to improve the ability of an automatedinsulin delivery system to automatically respond to large glycemic disturbances such as eating or exercise, using only sensor measurements without user input. In some embodiments, a Hidden Markov Model (HMM) is used when an automated insulin delivery algorithm is running on an individual to calculate the probability that various possible glycemic disturbances are currently occurring (i.e., eating starting, eating ending, aerobic exercise, anaeoribic exercise, etc.) given signals such as continuous glucose monitor measurements or heart rate monitors. These probabilities are then used as input to calculate current insulinPATENT APPLICATION delivery. The optimal insulin delivery for each glycemic disturbance can be calculated before the algorithm runs using a dynamic program that contains all or part of the Markov Model that is used at algorithm run time for glycemic disturbance detection. In various embodiments, the disclosed method can be applied to any infusion pump for delivering a dose of medicament.

[0035] Figure 1 depicts an embodiment of a medical device according to the disclosure. Inthis embodiment, the medical device is configured as a pump 12. Pump 12 may be an infusion pump that includes a pumping or delivery mechanism and reservoir for delivering medicament to a patient and an output / display 44. The output / display 44 may include an interactive and / or touch sensitive screen 46 having an input device such as, for example, a touch screen comprising a capacitive screen or a resistive screen. The pump 12 may additionally or instead include one or more of a keyboard, a microphone or other input devices known in the art for data entry, some or all of which may be separate from the display. The pump 12 may also include a capability to operatively couple to one or more other display devices such as a remote display, a remote-control device, a laptop computer, personal computer, tablet computer, a mobile communication device such as a smartphone, a wearable electronic watch or electronic health or fitness monitor, or personal digital assistant (PDA), a CGM display etc.

[0036] In one embodiment, the medical device can be an ambulatory insulin pumpconfigured to deliver insulin to a patient. Further details regarding such pump devices can be found in U.S. Patent No.8,287,495, which is incorporated herein by reference in its entirety. In other embodiments, the medical device can be an infusion pump configured to deliver one or more additional or other medicaments to a patient.

[0037] Figure 2 illustrates a block diagram of some of the features that can be used withembodiments, including features that may be incorporated within the housing 26 of a medical device such as a pump 12. The pump 12 can include a processor 42 that controls the overall functions of the device. The infusion pump 12 may also include, e.g., a memory device 30, a transmitter / receiver 32, an alarm 34, a speaker 36, a clock / timer 38, an input device 40, a user interface suitable for accepting input and commands from a user such as a caregiver or patient, a drive mechanism 48, an estimator device 52 and a microphone (not pictured). One embodiment of a user interface is a graphical user interface (GUI) 60 having a touch sensitive screen 46 with input capability. In some embodiments, the processor 42 may communicate with one or more other processors within the pump 12 and / or one or more processors of other devices, for example, a CGM, display device, smartphone, etc. through the transmitter / receiver. The processor 42 may also include programming that may allow the processor to receivePATENT APPLICATION signals and / or other data from an input device, such as a sensor that may sense pressure, temperature, or other parameters.

[0038] Figures 3A-3B depict another pump system including a pump 102 that can be usedwith embodiments. Drive unit 118 of pump 102 includes a drive mechanism 12 that mates with a recess in disposable cartridge 116 of pump 102 to attach the cartridge 116 to the drive unit 118. Pump system 100 can further include an infusion set 145 having a connector 154 that connects to a connector 152 attached to pump 102 with tubing 153. Tubing 144 extends to a site connector 146 that can attach or be pre-connected to a cannula and / or infusion needle that punctures the patient’s skin at the infusion site to deliver medicament from the pump 102 to the patient via infusion set 145. In some embodiments, pump can include a user input button 172 and an indicator light 174 to provide feedback to the user.

[0039] In one embodiment, pump 102 includes a processor that controls operations of thepump and, in some embodiments, may receive commands from a separate device for control of operations of the pump. Such a separate device can include, for example, a dedicated remote control or a smartphone or other consumer electronic device executing an application configured to enable the device to transmit operating commands to the processor of pump 102. In some embodiments, processor can also transmit information to one or more separate devices, such as information pertaining to device parameters, alarms, reminders, pump status, etc. In one embodiment, pump 102 does not include a display but may include one or more indicator lights 174 and / or one or more input buttons 172. Pump 102 can also incorporate any or all of the features described with respect to pump 12 in Figure 1. Further details regarding such pumps can be found in U.S. Patent No. 10,279,106 and U.S. Patent Publication Nos. 2016 / 0339172 and 2017 / 0049957, each of which is hereby incorporated herein by reference in its entirety.

[0040] In some embodiments, the pump 12 or 102 can interface directly or indirectly (via,e.g., a smartphone or other device) with a glucose meter, such as a blood glucose meter (BGM) or a CGM. Referring to Figure 4, an exemplary CGM system 100 according to an embodiment of the present invention is shown (other CGM systems can be used). The illustrated CGM system includes a sensor 101 affixed to a patient 104 that can be associated with the insulin infusion device in a CGM-pump system. The sensor 101 includes a sensor probe 106 configured to be inserted to a point below the dermal layer (skin) of the patient 104. The sensor probe 106 is therefore exposed to the patient’s interstitial fluid or plasma beneath the skin and reacts with that interstitial fluid to produce a signal that can be associated with the patient’s blood glucose level. The sensor 101 includes a sensor body 108 that transmits data associatedPATENT APPLICATION with the interstitial fluid to which the sensor probe 106 is exposed. The data may be transmitted from the sensor 101 to the glucose monitoring system receiver 100 via a wireless transmitter, such as a near field communication (NFC) radio frequency (RF) transmitter or a transmitter operating according to a “Wi-Fi” or Bluetooth® protocol, Bluetooth® low energy protocol or the like, or the data may be transmitted via a wire connector from the sensor 101 to the monitoring system 100. Transmission of sensor data to the glucose monitoring system receiver by wireless or wired connection is represented in Figure 4 by the arrow line 112. Further detail regarding such systems and definitions of related terms can be found in, e.g., U.S. Patent Nos. 8,311,749, 7,711,402 and 7,497,827, as well as PCT Patent App. No. PCT / US23 / 82084, each of which is hereby incorporated by reference in its entirety.

[0041] In an embodiment of a pump-CGM system having a pump 12, 102 thatcommunicates with a CGM and that integrates CGM data and pump data as described herein, the CGM can automatically transmit the glucose data to the pump. The pump can then automatically determine therapy parameters and deliver medicament based on the data. Such an automatic pump-CGM system for insulin delivery can be referred to as an automated insulin delivery (AID) or an artificial pancreas system that provides closed-loop therapy to the patient to approximate or even mimic the natural functions of a healthy pancreas. In such a system, insulin doses are calculated based on the CGM readings (that may or may not be automatically transmitted to the pump) and are automatically delivered to the patient at least in part based on the CGM reading(s). In various embodiments, doses can be delivered as automated correction boluses and / or automated increases or decreases to a basal rate. Insulin doses can also be administered based on current glucose levels and / or predicted future glucoses levels based on current and past glucose levels.

[0042] For example, if the CGM indicates that the user has a high blood glucose level orhyperglycemia, the system can automatically calculate an insulin dose necessary to reduce the user’s blood glucose level below a threshold level or to a target level and automatically deliver the dose. If the CGM data indicates that the user has a low blood glucose level or hypoglycemia, the system can, for example, automatically reduce a basal rate and / or make other suggestions as may be appropriate to address the hypoglycemic condition. As with other parameters related to therapy, such thresholds and target values can be stored in memory located in the pump and the pump processor can periodically and / or continually execute instructions for a checking function that accesses these data in memory, compares them with data received from the CGM and acts accordingly to adjust therapy. The complexity of the algorithm used to calculate thePATENT APPLICATION insulin doses is therefore limited by the capabilities of the pump processor, memory, battery, etc.

[0043] In some embodiments, systems include an infusion pump, which can receive dosingfunctions from an electronic device or a neural network, select one of the dosing functions based on characteristics of a user, deliver a dose of medicament to the user according to the selected dosing function. The characteristics may include a basal rate, a correction factor, or a matching factor estimated based on the basal rate and the correction factor. Further detail regarding such systems and definitions of related terms can be found in PCT Patent App. No. PCT / US23 / 82084, which is hereby incorporated by reference in its entirety.

[0044] In some embodiments, the system computes an insulin dosing target for a user basedon not only the data from the CGM, but also an activity state of the user. For example, given a same measurement result from the CGM indicating a same blood glucose level, the system can compute a first insulin dosing target if the user just starts to eat (at a first state), and compute a second insulin dosing target if the user is about to finish eating (at a second state). In some examples, the first insulin dosing target may be higher than the second insulin dosing target. This is because the user at the first state will receive more glucose soon, but the user at the second state will stop receiving glucose soon. In addition, it takes some time for dosed insulin to take effect in the user’s body.

[0045] Embodiments disclosed herein employ a dynamic and interoperable automatedglycemic control algorithm that estimates probabilities that the user is at different activity states at a given time step, and computes an insulin dosing target based on these probabilities, data from a glucose monitor, and predetermined optimal dosing target for each activity state. Referring to Figure 5, a block diagram of a system 200 for employing the automated glycemic control algorithm of the present disclosure is depicted according to embodiments. System 200 implements the automated glycemic control algorithm across a workstation 202 and an infusion pump 204 to provide optimized insulin delivery based on multiple processes (P0, P1, P2, P3, P4 and P5).

[0046] At P0, a dosing function (or dosing policy) generation process is performed atworkstation 202 (which is or may include a “dosing function device” as described in this disclosure). In some embodiments, the policy generation process incorporates the combination of a model-predictive-control (MPC) algorithm and dynamic programming. The general concept behind an MPC algorithm is use of multiple control variables with multiple targets to simulate thousands (or even millions) of scenarios for each decision point. In some examples, a user is in one of multiple activity states at any given time, and the user is capable ofPATENT APPLICATION transitioning among the activity states during a certain time period, e.g., during daily living. For example, the activity states comprise at least four eating states of the user: not eating, eating start, eating middle, and eating finish. A predictive model can be used to simulate a hypothetical state transition of a user from one activity state to another, e.g., from not eating to eating start, and simulate effects of different dosing decisions over time for the user. For each respective one of the activity states, a dosing function can be determined based on the predictive model, to represent optimal insulin dosing targets given different glucose influx levels of the user.

[0047] In some embodiments, each dosing function determined for a given activity state isalso optimized and customized for the user. For example, the dosing function, for a given activity state, is determined based on: a selection or combination of multiple candidate dosing functions utilizing one or more characteristics of the user. In some examples, the one or more user characteristics comprise a target blood glucose, a basal rate, a correction factor, and a carb ratio. In some examples, a matching factor is estimated by multiplying the basal rate by the correction factor. The selection or combination (e.g., interpolation) of the candidate dosing functions may be based on the matching factor and a target glucose.

[0048] In some embodiments, each of the candidate dosing functions represents acumulative cost associated with each termination state. Each termination state represents an outcome following a sequence of dosing decisions over a finite time horizon, e.g., based on simulation results from an MPC algorithm. Each termination state has an associated cumulative cost by applying a cost function to each dose in the sequence of dosing decisions. In some embodiments, there are five meal states: meal state 0 (e.g., not eating), meal state 1 (e.g., eating start), meal state 2 (e.g., eating fast), meal state 3 (e.g., eating slow), meal state 4 (e.g., eating finish). The workstation 202 can perform a first simulation based on MPC algorithm, based on meal state 0, to select an optimal candidate dosing function, or generate the optimal candidate dosing function based on a weighted average of two or more candidate dosing functions. After the optimal dosing function is determined for meal state 0, the workstation 202 can perform a second simulation based on MPC algorithm, to determine other optimal dosing functions for other meal states, e.g., based on a shift from the optimal dosing function determined for meal state 0. In other embodiments, the workstation 202 can perform a single simulation to generate all optimal dosing functions for all meal states.

[0049] The P0 logic resides on workstation 202 and can generate the policies offline. Thesepolicies can then be loaded onto the pump during the manufacturing process or through a programmer. In some embodiments, P0 can be implemented on any computing devices capablePATENT APPLICATION of remotely or locally handling the associated computational overhead, such as a smartphone, a tablet computer, a laptop computer, a desktop computer, or a high-end workstation.

[0050] At P1, the infusion pump logic receives the dosing policies (e.g., dosing functions)from the workstation 202, and generates a combined dosing policy (or dosing function) based on the dosing policies generated by the workstation 202. The combined dosing policy may be a blend of all dosing policies generated by the workstation 202, based on state probabilities generated at P3 and computed based on filtered measurement data at P2.

[0051] At P2, measurement data of the user from a monitor, e.g., a CGM, is filtered. Insome embodiments, an estimated value is computed based on applying a filter on the measurement result on the user from the monitor. In some examples, the filter may be a Kalman filter, and can be updated based on a dose of medicament delivered to the user at a previous time step. In some embodiments, the estimated value represents an estimated glucose influx for the user at a current time step and will be used at P3 for computing state probabilities.

[0052] At P3, state scores for different activity states are computed. Each of the state scoresrepresents a probability that the user is in a respective one of the activity states at a given time. In some embodiments, the state scores (representing or proportional to state probabilities) are computed based on a hidden Markov model (HMM). The HMM models a system with a combination of a Markov Model and a collection of probability distributions over some space of signals (e.g., a measurement of blood glucose). A Markov Model models a system that can change over time as being in one of multiple several discrete states at a given time, and time is divided into discrete steps following the model. The state can stochastically change over time. The probability of transitioning between any pair of discrete states is given by a transition matrix.

[0053] In some embodiments, the transition matrix is computed based on a simulation ofmultiple patients’ daily living and an optimization performed before the run time of the algorithm at the system 200. In some embodiments, the transition matrix is fixed for all users. In some embodiments, the transition matrix is updated for each specific user. In some embodiments, the transition matrix is optimized to ensure that there is no sinking state where a user will not exit once getting in that state. In some embodiments, the transition matrix is optimized or updated based on historical insulin delivery amounts. For example, the state transition may be impacted by the insulin delivery happened several time steps ago. In some embodiments, the filter applied at P2 is also affected by the state transition and / or the insulin delivery happened several time steps ago.PATENT APPLICATION

[0054] Figure 7 illustrates a hidden Markov model (HMM) used for estimating meal statesof a user, according to various embodiments of the present disclosure. In the example shown in FIG.7, a Markov Model modeling eating includes 4 states: not eating (ne), eating start (es), eating middle (em), and eating finish (ef). In Figure 7, each possible state is represented as a circle, possible transitions between them are represented as arrows, and each arrow is labeled with the probability of that transition per time step. For example, a transition from not eating (ne) 710 to eating start (es) 720 has a probability of p(ne->es), and a transition from eating middle (em) 730 to eating finish (ef) 740 has a probability of p(em->ef). There is no arrow between not eating (ne) 710 and eating middle (em) 730, because of a natural logic of eating process. In other words, the probability of transitioning between not eating (ne) 710 and eating middle (em) 730 is zero. All of these transition probabilities of transitioning between any pair of the meal states are included in a transition matrix T, which will be used to compute and / or update state probabilities.

[0055] In some embodiments, each state is associated with a conditional probabilitydistribution of receiving a particular value of signals given the discrete state at a time step. These probability distributions are called emissions probabilities. Referring back to Figure 5, following the Hidden Markov Model, each of these states at P3 is associated with the probability of observing a value of estimated glucose influx ^^^^^^^^, which may be estimated from the glucose measurements of a continuous glucose monitor at P2. As discussed above, the glucose influx ^^^^^^^^may be estimated via an application of a metabolic model and Kalman Filter. In some embodiments, the glucose influx ^^^^^^^^is clipped based on a maximum value and a minimum value, to ensure the glucose influx ^^^^^^^^does not exceed the range between the minimum value and the maximum value. This can help preventing extreme sensor error from the CGM, which would cause a sudden deviate of dose amount. The extreme sensor error or sudden glucose measurement change may be due to a user leaning on the CGM. In some embodiments, the glucose influx ^^^^^^^^is also normalized before being used at P1.

[0056] In general, the estimated glucose influx ^^^^^^^^ is expected to be lowest when noteating and highest in the middle of eating. As such, an example set of emission probabilities is shown in Figure 8, where dependence of emission probability distributions 810, 820, 830 on glucose influx of a user at different meal states, according to various embodiments of the present disclosure. For example, the emission probability distribution 810 for not eating state has a smallest mean value among the three emission probability distributions 810, 820, 830, because the estimated glucose influx ^^^^^^^^is expected to be lowest when not eating.PATENT APPLICATION

[0057] Referring back to Figure 5, the HMM can be used for meal-like glycemicdisturbance detection. In some embodiments, at P3, the infusion pump 204 runs an algorithm to track a vector of the probability of each of the eating states: [^^^^^^^^^^^^^^^^^^^^^^^^]. At initialization, this vector is

[1000] . At each time step the algorithm runs, ^^^^^^^^is calculated at P2, e.g., using Kalman filter on incoming CGM readings. Then at P3, emission probabilities of that particular value of ^^^^^^^^is calculated from the distributions for different meal states. In this example, these values of emission probabilities given four meal states are represented as: ^^(^^^^^^^^|^^^^),^^(^^^^^^^^|^^^^),^^(^^^^^^^^|^^^^), and^^(^^^^^^^^|^^^^), respectively. In some embodiments,Bayes rule can be used to calculate updated values of the probability of each eating state ^^ (^^^^|^^^^^^^^), ^^(^^^^|^^^^^^^^), ^^(^^^^|^^^^^^^^), and ^^(^^^^|^^^^^^^^), given the measurement and estimated value ^^^^^^^^, based on an update of state vector [^^^^^^^^^^^^^^^^^^^^^^^^]. In some embodiments, the probability for receiving the estimated value ^^^^^^^^, ^^(^^^^^^^^), can be obtained from apriori measurement data and used to calculate the conditional probabilities ^^(^^^^|^^^^^^^^), ^^(^^^^|^^^^^^^^),^^(^^^^|^^^^^^^^), and^^(^^^^|^^^^^^^^). In some embodiments, these conditional probabilities will be utilized for weighting different states. As such, the probability for receiving the estimated value ^^^^^^^^, ^^(^^^^^^^^), can be treated as a normalization factor in the calculation of the conditional probabilities. In some embodiments, the weights to be used for combining different dosing policies or dosing functions will be normalized weights summing to one, and proportional tothe conditional probabilities ^^(^^^^|^^^^^^^^), ^^(^^^^|^^^^^^^^), ^^(^^^^|^^^^^^^^), and ^^(^^^^|^^^^^^^^). In someembodiments, the state vector[^^^^^^ ^^^^^^ ^^^^^^ ^^^^^^]at each time step is updated based on thetransition probabilities of the Markov model and the state vector[^^^^^^ ^^^^^^ ^^^^^^ ^^^^^^]at apreceding time step.

[0058] The updated vector of the probability of each of the eating states ^^(^^^^|^^^^^^^^), ^^(^^^^|^^^^^^^^),^^(^^^^|^^^^^^^^), and^^(^^^^|^^^^^^^^), serves as an input to blending dosing policies forcalculating the optimal insulin delivery at P1.

[0059] In some embodiments, the optimal insulin delivery given an eating state has beencalculated before the algorithm runs on the user at the infusion pump 204, following the dosing policies generated by the workstation 202. As discussed above, the dosing policies may be generated using a dynamic program to solve for optimal delivery given a metabolic predictive model. In some examples, the metabolic predictive model also includes a Markov Model similar to that shown in Figure 7. In some examples, the Markov Model, used for simulation and determining the optimal dosing policies for different meal states at P0 by the workstationPATENT APPLICATION 202, is the same as or a part of the Markov Model used at P3 by the infusion pump 204 for computing the probabilities of the various eating states at run time.

[0060] In some embodiments, the actual insulin to deliver given the vector of probabilitiesof each eating state may be calculated by a weighted average of the calculated optimal insulin delivery for each eating state. The weights are given by the probability of each eating state computed at P3. In some examples, the dosing policies generated by the workstation 202 for different meal states can be blended using a weighted average of these dosing policies at P1, where weights are given by the state probabilities computed at P3.

[0061] In some embodiments, at P1, the automated glycemic control algorithm also usesthe blended policy and the estimated glucose influx of the user (e.g., based on filtered data from the CGM at P2) to compute an optimum dosing target for each time step, e.g., each five-minute period. The optimum dosing target will be sent to P5 for determining an optimal amount of insulin dose and delivering the optimal amount of insulin dose to the user according to the combined dosing function or blended policy.

[0062] In some embodiments, this P5 logic includes modules for estimating anunmetabolized insulin (UMI), approximating insulin onboard (IOB), and generating glucose predictions. In some embodiments, P5 considers all insulin active in the body based on insulin estimate at P4. In some embodiments, the dosed amount at each time step from P5 is provided to P4 as feedback. The P4 logic is to estimate a current insulin amount (or in general an amount of medicament being dosed) in the body of the user based on: previous doses performed at P5. The P5 logic can compute the optimal amount of insulin dose based on: the estimated current insulin amount from P4 and the optimum insulin dosing target from P1, for the current time step. In some embodiments, an estimated IOB mean is computed at P4 and provided to P2 for filtering the estimated glucose influx.

[0063] As such, the operations of the infusion pump 204, including policy blending, statescore computations for different activity states, and optimal dosing target determination, are all performed automatically without any input from the user. In some embodiments, the disclosed system 200 can be applied to dose any medicament based on a blended policy across different activity states of a user. In some embodiments, the activity states may include meal states, exercise states, or both. For example, while the user is not eating, the not eating state may be further divided into different exercise states: e.g., not exercising, exercising start, exercising heavily, exercising slightly, exercising finish.

[0064] Contemplated herein is the employment of a hidden Markov model to detect theprobability of a mixture of different glycemic disturbances (e.g., different meal states, differentPATENT APPLICATION exercise states) and respond appropriately given that these disturbances may stochastically transition into each other. The disturbances are not modeled individually, but as a collection.

[0065] In some embodiments, the optimal insulin delivery for each glycemic disturbancecan be computed beforehand and stored in a compressed form using dynamic programming, making it suitable for deploying an algorithm in embedded contexts. For example, the infusion pump 204 further comprises a memory. Prior to delivering the dose of the medicament to the user, the infusion pump 204 can store the dosing functions or policies (obtained from the workstation 202) in the memory. In some embodiments, an optimal number of activity states is also predetermined based on a simulation run beforehand.

[0066] Referring now to Figure 6, a flow chart of a method 300 for delivering optimalinsulin doses based on blended policies is depicted according to an embodiment. In some embodiments, method 300 can be implemented through the system 200.

[0067] It is contemplated that the processing logic implemented at step 302 can be ran ona workstation independent from an infusion pump. The computations can be performed on a computing system having computational capabilities greater than those of an ambulatory infusion pump. At step 302, dosing policies are determined based on a simulation with a predictive model. In some embodiments, the predictive model utilizes a simulation tree represented using MPC, where the representation may implicitly contain the same possible paths as the simulation tree and may be in a tabular form such as a look-up table, which is initially empty. In some embodiments, dynamic programming is used to reduce the burden of MPC by removing unnecessary calculations. A backward pass may be performed on the implicit representation of the simulation tree in the table to fill out the empty table which becomes the policies / lookup tables. Each of the dosing policies can represent dosing targets associated with a respective one of activity states (e.g., meal states and / or exercise states). A user is in one of the activity states at any given time, and is capable of transitioning among the activity states during a certain time period, e.g., during daily living.

[0068] These dosing policies can then be stored in the memory of the pump and / or remotecontrol at step 304. At step 306, state probabilities are computed, where each state probability represents a probability that the user is in a respective one of the activity states at a given time step. In some embodiments, the state probabilities are computed by: (1) building a hidden Markov model (HMM) that models the user who changes over time steps as being in one of the activity states at a given time step; (2) determining, for the HMM, a transition matrix including transition probabilities of transitioning between any pair of the activity states; (3) determining preceding state probabilities at a preceding time step, each of the preceding statePATENT APPLICATION probabilities representing a probability that the user was in a respective one of the activity states at the preceding time step; (4) computing an estimated value based on a measurement on the user (e.g., an estimated glucose influx); (5) determining emission probabilities at a current time step, each of the emission probabilities is a conditional probability of obtaining the estimated value given that the user is in a respective one of the activity states at the current time step; and (6) computing the state probabilities at the current time step based on: the transition matrix, the preceding state probabilities at the preceding time step, and the emission probabilities at the current time step.

[0069] At step 308, the dosing policies can be obtained from the memory and blendedbased on the state probabilities. For example, a blended policy can be computed based on a weighted average of the dosing policies, where the weight for each dosing policy corresponding to an activity state is the state probability computed for that activity state. In some embodiments, the dosing policies are blended by: (1) representing each of the dosing policies based on a respective coefficient set of coefficients corresponding to a respective one of the activity states, where each coefficient in the respective coefficient set corresponds to an index; (2) computing weighted coefficients each for a respective index, where each of the weighted coefficients is computed based on a weighted average of coefficients corresponding to the respective index in all of the coefficient sets, with corresponding weights being the state probabilities; and (3) generating a combined or blended dosing policy based on the weighted coefficients.

[0070] At step 310, an estimated value (e.g., an estimated glucose influx) for a user iscomputed based on measurement data from a sensor, e.g., a CGM, for each dosing interval (e.g., every five minutes). In some embodiments, the measurement data is filtered to generate a distribution of glucose influx at a given time step. In that case, a mean of the distribution may be used as the estimated glucose influx. At step 312, embedded logic in the pump and / or remote control can use the blended policy to determine appropriate doses (e.g., an optimal insulin dose) for each dosing interval (e.g., every five minutes), based on the estimated glucose influx. This implementation effectively simplifies real-time computation to a lookup operation by the infusion pump. Finally, at step 314, the insulin pump delivers the determined insulin dose.

[0071] Steps 310 through 314 are performed for each dosing interval or time step and canbe repeated as necessary to enable the automated glycemic control algorithm to operate ad infinitum without running out of policies. For example, at a next time step, the infusion pump can: (1) compute new state probabilities based on the state probabilities at the current time step and a matrix of transition probabilities for transitioning between any pair of the activity states,PATENT APPLICATION each of the new state probabilities representing a probability that the user is in a respective one of the activity states at the next time step; (2) generate a new blended dosing policy based on the dosing policies weighted by the new state probabilities; and (3) deliver a new insulin dose to the user according to the new blended dosing policy, based on an updated estimate of the glucose influx of the user.

[0072] It should be understood that the individual operations used in the methods of thepresent teachings may be performed in any order and / or simultaneously, as long as the teaching remains operable. Furthermore, it should be understood that the apparatus and methods of the present teachings can include any number, or all, of the described embodiments, as long as the teaching remains operable.

[0073] In some embodiments, an optimal dosing function or policy for each meal state isaffected by the optimal dosing for the other meal states based on the transition probabilities from one meal state to another. In some embodiments, an optimal insulin value is affected by the glucose and insulin variables in the future. The Markov Model can make the future trajectories of glucose and insulin variables stochastic instead of deterministic. In some embodiments, a mean future cost, rather than a deterministic cost, is optimized over the future.

[0074] In some embodiments, policies are lookup tables generated by the automatedglycemic control algorithm policy generator. A typical policy (given an activity state) is depicted in Figure 9A according to an embodiment and elements of that policy are identified in Figures 9B-9D. Each policy defines the relationship between the estimated glucose and the UMI-Target over a range. In the embodiment depicted in Figures 9A-9D, the policy defines the range of 40 mg / dL to 600 mg / dL.

[0075] Referring to Figure 9B, each policy includes three segments: a flat portion, a quickrise portion, and a slow rise portion. In the flat portion, UMI-Target should be zero as this represents flat glucose. In the quick rise portion, UMI-Target should rise approximately linearly (sometimes following a sigmoid) with predicted glucose, and the largest y-value should occur at the predicted glucose target. In the slow rise portion, UMI-Target shall have a positive first derivative and a negative second derivative. Each policy has both (i) a maximum dose and (ii) end of flat glucose value, as depicted in Figure 9C. The impact of adjusting automated glycemic control algorithm parameters is illustrated in Figure 9D. Adjusting automated glycemic control algorithm Target Glucose tends to translate the policy to the right or left. Adjusting automated glycemic control algorithm Basal tends to scale the policy up and down. Adjusting automated glycemic control algorithm Correction Factor tends to rotate the policy Clockwise (CW) or Counter-Clockwise (CCW).PATENT APPLICATION

[0076] Because each policy can be personalized, each user will have an optimum policygiven an activity state. Figure 10 is a graph showing different policies corresponding to different activity states, according to various embodiments of the present disclosure. In the example shown in Figure 10, dosing policies for five meal states (MS0, MS1, MS2, MS3, MS4) are illustrated. As shown in Figure 10, the five dosing policies share the same flat portion between E1 and E2, have different quick rise portions between E2 and E3, and have different slow rise portions between E3 and E4. In some embodiments, the slopes of the quick rise portions of the five dosing policies can be blended based on a weighted average of these slopes, with weights being the corresponding state probabilities computed for the five meal states at a given time step and given an estimated glucose influx, to generate a blended slope. In some embodiments, each slow rise portion is modelled or fit by a same-order polynomial function with coefficients. Those slow rise portions are blended based on a weighted average of these coefficients, with weights being the corresponding state probabilities computed for the five meal states at a given time step and given an estimated glucose influx, to generate a set of blended coefficients. For example, each slow rise portion can be modelled as a fifth-order polynomial function with six coefficients. A weighted average of the six coefficients of all dosing policies can be computed to generate a set of blended six coefficients. The blended policy is then represented by: the same flat portion between E1 and E2, a quick rise portion with the blended slope between E2 and E3, and a slow rise portion with the blended coefficients between E3 and E4.

[0077] Figure 11 illustrates optimal all-IOB (“aIOB” or sometimes referred to as “AOB”)target predictions 2102, according to various embodiments of the present disclosure. The graph 1100 illustrates a dosing function, with an aIOB target 2102 as a function of estimated glucose. For example, a system (e.g., the system 200) can determine an optimal aIOB for a patient based in part on an estimated blood glucose level of the patient, given an activity state.

[0078] Figure 12 is a graph 1200 showing different dosing functions corresponding todifferent meal states, according to various embodiments of the present disclosure. In the example shown in Figure 12, dosing functions for five meal states (MS0, MS1, MS2, MS3, MS4) are illustrated. The dosing functions in Figure 12 are similar to the dosing functions in Figure 10, except that each dosing function in Figure 12 represents an aIOB target rather than UMI-Target as in Figure 10. The blending method described above for Figure 10 can be applied to blend these dosing functions in Figure 12 as well.

[0079] Figure 13 illustrates a process 1300 for blending different dosing functions fordifferent meal states at a given time step, according to various embodiments of the presentPATENT APPLICATION disclosure. In some embodiments, the process 1300 is implemented as part of the process P1 by the infusion pump 204 discussed above with respect to Figure 5.

[0080] As shown in Figure 13, the process 1300 starts at step 1310, where it is determinedwhether the user is sleeping. If so, the process goes to step 1330 to determine that the blended dosing function is the same as the dosing function for meal state 0. This is because it is assumed that the user is not eating (meal state 0) when the user is sleeping. As such, all coefficients for the blended dosing function are the same as the coefficients of the dosing function for meal state 0.

[0081] Otherwise, if it is determined the user is not sleeping at step 1310, the process goesto step 1320 to determine whether the user is exercising. If so, the process goes to step 1330 to determine that the blended dosing function is the same as the dosing function for meal state 0. This is because it is assumed that the user is not eating (meal state 0) when the user is exercising. As such, all coefficients for the blended dosing function are the same as the coefficients of the dosing function for meal state 0.

[0082] Otherwise, if it is determined the user is not exercising at step 1320, the processgoes to step 1340 to generate the blended dosing function based on all meal states. In this example, there are total S meal states (meal states 0 to S-1), with meal state 0 representing not eating. As such, each coefficient for the blended dosing function is a blend or mix of the corresponding coefficients of dosing functions for all meal states, with weights being the corresponding state probabilities (mp_0, mp_1 … mp_(S-1)) computed for the meal states at a given time step and given an estimated glucose influx.

[0083] Figure 14 is a flow chart of a method 1400 for delivering a dose of medicamentbased on a combined dosing function, according to various embodiments of the present disclosure. In some embodiments, the method 1400 is implemented by an infusion pump, such as the infusion pump 204 discussed above with respect to Figure 5.

[0084] For illustrative purposes, the following discussion refers to “a processor” asperforming the operations of the method 1400. This processor may be a processor of the aforenoted infusion pump. Additionally, the processor may be a processor of a server, a mobile device, or any other electronic device discussed herein or otherwise known in the art. Moreover, it is noted that some of the operations of the method 1400 can be performed in an order other than the serial order suggested by the Figure 14 and the following discussion. Likewise, some of the operations can be performed in parallel, and, in some embodiments, some of the operations need not be performed whatsoever.PATENT APPLICATION

[0085] In the illustrated embodiment, the processor receives (1410) a set of dosingfunctions from a dosing function device (e.g., the workstation 202 in Figure 5), each of the dosing functions representing dosing targets associated with a respective one of activity states. A user is in one of the activity states at any given time and is capable of transitioning among the activity states during a certain time period, e.g., during daily living. For example, the dosing functions may be generated at P0 in Figure 5.

[0086] After receiving the dosing functions, the processor computes (1420) state scores,each of the state scores representing a probability that the user is in a respective one of the activity states. For example, the state scores may be computed as the state probabilities estimated at P3 in Figure 5.

[0087] Then, the processor generates (1430) a combined dosing function based on thedosing functions and the state scores. As discussed above, the combined dosing function may be a weighted average of the dosing functions for different activity states, with the weights being the state scores for the corresponding activity states. For example, the combined dosing function may be generated at P1 in Figure 5.

[0088] Then the processor causes (1440) an infusion pump to deliver a dose of medicamentto the user according to the combined dosing function. For example, the dose of medicament may be an optimal insulin dose delivered at P5 in Figure 5. The optimal insulin dose is computed based on an insulin dosing target, which is in turn determined based on the combined dosing function and an estimated glucose influx for the user.

[0089] In some embodiments, the disclosed systems and method can be applied to dose anymedicament based on a blended policy across different activity states of a user. In some embodiments, the activity states include different exercise states, and the system can receive data from a heart rate monitor to compute an estimated heart rate of the user at a given time step. The state probabilities for different exercise states will be computed based on the estimated heart rate.

[0090] Activity states may include meal states, exercise states, or both. For example, whilethe user is not eating, the not eating state may be further divided into different exercise states: e.g., not exercising, exercising start, exercising heavily, exercising slightly, exercising finish.

[0091] Computing and other devices discussed herein can include memory. Memory cancomprise volatile or non-volatile memory as required by the coupled computing device or processor to not only provide space to execute the instructions or algorithms, but to provide the space to store the instructions themselves. In one embodiment, volatile memory can include random access memory (RAM), dynamic random access memory (DRAM), or static randomPATENT APPLICATION access memory (SRAM), for example. In one embodiment, non-volatile memory can include read-only memory, flash memory, ferroelectric RAM, hard disk, floppy disk, magnetic tape, or optical disc storage, for example. The foregoing lists in no way limit the type of memory that can be used, as these embodiments are given only by way of example and are not intended to limit the scope of the disclosure.

[0092] In one embodiment, the system or components thereof can comprise or includevarious modules or engines, each of which is constructed, programmed, configured, or otherwise adapted to autonomously carry out a function or set of functions. The term “engine” as used herein is defined as a real-world device, component, or arrangement of components implemented using hardware, such as by an application specific integrated circuit (ASIC) or field programmable gate array (FPGA), for example, or as a combination of hardware and software, such as by a microprocessor system and a set of program instructions that adapt the engine to implement the particular functionality, which (while being executed) transform the microprocessor system into a special-purpose device.

[0093] An engine can also be implemented as a combination of the two, with certainfunctions facilitated by hardware alone, and other functions facilitated by a combination of hardware and software. In certain implementations, at least a portion, and in some cases, all, of an engine can be executed on the processor(s) of one or more computing platforms that are made up of hardware (e.g., one or more processors, data storage devices such as memory or drive storage, input / output facilities such as network interface devices, video devices, keyboard, mouse or touchscreen devices, etc.) that execute an operating system, system programs, and application programs, while also implementing the engine using multitasking, multithreading, distributed (e.g., cluster, peer-peer, cloud, etc.) processing where appropriate, or other such techniques. Accordingly, each engine can be realized in a variety of physically realizable configurations and should generally not be limited to any particular implementation exemplified herein, unless such limitations are expressly called out. In addition, an engine can itself be composed of more than one sub-engine, each of which can be regarded as an engine in its own right. Moreover, in the embodiments described herein, each of the various engines corresponds to a defined autonomous functionality; however, it should be understood that in other contemplated embodiments, each functionality can be distributed to more than one engine. For example, in an embodiment, each of the processes depicted in Figure 5 could be implemented within engines as described above. Likewise, in other contemplated embodiments, multiple defined functionalities may be implemented by a single engine thatPATENT APPLICATION performs those multiple functions, possibly alongside other functions, or distributed differently among a set of engines than specifically illustrated in the examples herein.

[0094] Various embodiments of systems, devices, and methods have been described herein.These embodiments are given only by way of example and are not intended to limit the scope of the claimed inventions. It should be appreciated, moreover, that the various features of the embodiments that have been described may be combined in various ways to produce numerous additional embodiments. Moreover, while various materials, dimensions, shapes, configurations and locations, etc. have been described for use with disclosed embodiments, others besides those disclosed may be utilized without exceeding the scope of the claimed inventions.

[0095] Persons of ordinary skill in the relevant arts will recognize that embodiments maycomprise fewer features than illustrated in any individual embodiment described above. The embodiments described herein are not meant to be an exhaustive presentation of the ways in which the various features may be combined. Accordingly, the embodiments are not mutually exclusive combinations of features; rather, embodiments can comprise a combination of different individual features selected from different individual embodiments, as understood by persons of ordinary skill in the art. Moreover, elements described with respect to one embodiment can be implemented in other embodiments even when not described in such embodiments unless otherwise noted. Although a dependent claim may refer in the claims to a specific combination with one or more other claims, other embodiments can also include a combination of the dependent claim with the subject matter of each other dependent claim or a combination of one or more features with other dependent or independent claims. Such combinations are proposed herein unless it is stated that a specific combination is not intended. Furthermore, it is intended also to include features of a claim in any other independent claim even if this claim is not directly made dependent to the independent claim.

[0096] Moreover, reference in the specification to “one embodiment,” “an embodiment,”or “some embodiments” means that a particular feature, structure, or characteristic, described in connection with the embodiment, is included in at least one embodiment of the teaching. The appearances of the phrase “in one embodiment” in various places in the specification are not necessarily all referring to the same embodiment.

[0097] Any incorporation by reference of documents above is limited such that no subjectmatter is incorporated that is contrary to the explicit disclosure herein. Any incorporation by reference of documents above is further limited such that no claims included in the documents are incorporated by reference herein. Any incorporation by reference of documents above isPATENT APPLICATION yet further limited such that any definitions provided in the documents are not incorporated by reference herein unless expressly included herein. For purposes of interpreting the claims, it is expressly intended that the provisions of Section 112, sixth paragraph of 35 U.S.C. are not to be invoked unless the specific terms “means for” or “step for” are recited in a claim.

Claims

PATENT APPLICATION CLAIMS What is claimed is:

1. An infusion pump comprising a processor and a non-transitory, computer-readablemedium storing instructions which, when executed by the processor, cause the infusion pump to: obtain a set of dosing functions, each of the dosing functions representing a dosing target associated with a respective one of activity states, wherein: a user is in one of the activity states at any given time, and the user is capable of transitioning among the activity states during a certain time period; determine state scores, each of the state scores representing a probability that the user is in a respective one of the activity states; determine a combined dosing function based on the dosing functions and the state scores; and deliver a dose of medicament to the user according to the combined dosing function.

2. The infusion pump of Claim 1, wherein:the activity states comprise at least four eating states of the user: not eating, eating start, eating middle, and eating finish; and the state scores for the activity states are computed without any input from the user.

3. The infusion pump of Claim 1, wherein the infusion pump is configured to determinethe state scores based on a hidden Markov model (HMM) that models the user who changes over time steps as being in one of the activity states at a given time step, and to determine the state scores, the infusion pump is configured to: determine, for the HMM, a transition matrix including transition probabilities of transitioning between any pair of the activity states; determine preceding state scores at a preceding time step, each of the preceding state scores representing a probability that the user was in a respective one of the activity states at the preceding time step; compute an estimated value based on a measurement on the user;PATENT APPLICATION determine emission probabilities at a current time step, each of the emission probabilities is a conditional probability of obtaining the estimated value given that the user is in a respective one of the activity states at the current time step; and compute the state scores at the current time step based on: the transition matrix, the preceding state scores at the preceding time step, and the emission probabilities at the current time step.

4. The infusion pump of Claim 3, wherein:the estimated value is computed based on applying a filter on a measurement result on the user from a monitor; and the filter is updated based on a dose of medicament delivered to the user at a previous time step.

5. The infusion pump of Claim 1, wherein to generate the combined dosing function, theinfusion pump is configured to: represent each of the dosing functions based on a respective coefficient set of coefficients corresponding to a respective one of the activity states, wherein each coefficient in the respective coefficient set corresponds to an index; compute weighted coefficients each for a respective index, wherein each of the weighted coefficients is computed based on a weighted average of coefficients corresponding to the respective index in all of the coefficient sets, with corresponding weights being the state scores; and generate the combined dosing function based on the weighted coefficients.

6. The infusion pump of Claim 1, wherein the infusion pump is further configured to, at anext time step after delivering the dose of the medicament to the user according to the combined dosing function: compute additional state scores based on the state scores and a matrix of transition probabilities for transitioning between any pair of the activity states, each of the additional state scores representing a probability that the user is in a respective one of the activity states at the next time step; generate an additional combined dosing function based on the dosing functions and the additional state scores; andPATENT APPLICATION deliver an additional dose of medicament to the user according to the additional combined dosing function.

7. The infusion pump of Claim 1, the infusion pump is further configured to:compute an estimated glucose influx of the user based on data from a continuous glucose monitor (CGM); determine an optimal amount of the dose of the medicament based on the combined dosing function and the estimated glucose influx; and deliver the dose of the medicament to the user comprises delivering the optimal amount of the dose according to the combined dosing function.

8. The infusion pump of Claim 1, further comprising a memory, wherein the infusionpump is further configured to, prior to delivering the dose of the medicament to the user, store the dosing functions in the memory.

9. A system for operation of an infusion pump, the system comprising:a dosing function device configured to: generate a predictive model to simulate effects of dosing decisions over time; determine activity states based on the predictive model, wherein: a user is in one of the activity states at any given time, and the user is capable of transitioning among the activity states during a certain time period; and determine dosing functions, each of the dosing functions representing a dosing target associated with a respective one of the activity states; and an infusion pump configured to: obtain the dosing functions from the dosing function device; determine state scores, each of the state scores representing a probability that the user is in a respective one of the activity states; determine a combined dosing function based on the dosing functions and the state scores; and deliver a dose of medicament to the user according to the combined dosing function.PATENT APPLICATION10. The system of Claim 9, wherein:the activity states comprise at least four eating states of the user: not eating, eating start, eating middle, and eating finish; and the state scores for the activity states are computed without any input from the user.

11. The system of Claim 9, wherein the infusion pump is configured to determine the statescores based on a hidden Markov model (HMM) that models the user who changes over time steps as being in one of the activity states at a given time step, and to determine the state scores, the infusion pump is configured to: determine, for the HMM, a transition matrix including transition probabilities of transitioning between any pair of the activity states; determine preceding state scores at a preceding time step, each of the preceding state scores representing a probability that the user was in a respective one of the activity states at the preceding time step; compute an estimated value based on a measurement on the user; determine emission probabilities at a current time step, each of the emission probabilities is a conditional probability of obtaining the estimated value given that the user is in a respective one of the activity states at the current time step; and compute the state scores at the current time step based on: the transition matrix, the preceding state scores at the preceding time step, and the emission probabilities at the current time step.

12. The system of Claim 11, wherein:the estimated value is computed based on applying a filter on a measurement result on the user from a monitor; and the filter is updated based on a dose of medicament delivered to the user at a previous time step.

13. The system of Claim 9, wherein to generate the combined dosing function, the infusionpump is configured to: represent each of the dosing functions based on a respective coefficient set of coefficients corresponding to a respective one of the activity states, wherein each coefficient in the respective coefficient set corresponds to an index;PATENT APPLICATION compute weighted coefficients each for a respective index, wherein each of the weighted coefficients is computed based on a weighted average of coefficients corresponding to the respective index in all of the coefficient sets, with corresponding weights being the state scores; and generate the combined dosing function based on the weighted coefficients.

14. The system of Claim 9, wherein the dosing function device is configured to:determine termination states, each of the termination states representing an outcome following a sequence of dosing decisions over a finite time horizon; apply a cost function to each dose in the sequence of dosing decisions such that each termination state has an associated cumulative cost; determine candidate dosing functions, each of the candidate dosing functions representing the cumulative cost associated with each termination state; obtain one or more characteristics of the user; and determine at least one of the dosing functions based on a selection or combination of the candidate dosing functions utilizing the one or more characteristics.

15. The system of Claim 9, wherein the one or more characteristics of the user comprise atarget blood glucose, a basal rate, a correction factor, and a carb ratio.

16. The system of Claim 9, wherein the infusion pump is further configured to, at a nexttime step after delivering the dose of the medicament to the user according to the combined dosing function: compute additional state scores based on the state scores and a matrix of transition probabilities for transitioning between any pair of the activity states, each of the additional state scores representing a probability that the user is in a respective one of the activity states at the next time step; generate an additional combined dosing function based on the dosing functions and the additional state scores; and deliver an additional dose of medicament to the user according to the additional combined dosing function.PATENT APPLICATION17. The system of Claim 9, further comprising a continuous glucose monitor (CGM), andwherein: the infusion pump is further configured to: compute an estimated glucose influx of the user based on data from the CGM, determine an optimal amount of the dose of the medicament based on the combined dosing function and the estimated glucose influx; and deliver the dose of the medicament to the user comprises delivering the optimal amount of the dose according to the combined dosing function.

18. The system of Claim 17, wherein:the infusion pump is further configured to (i) estimate an amount of UMI in the user, (ii) approximate an amount of insulin-on-board (IOB) in the user, or (iii) generate a glucose prediction for the user; and wherein determining the optimal amount of the dose is further based on (i) the estimated amount of UMI, (ii) the approximated amount of IOB, or (iii) the generated glucose prediction.

19. The system of Claim 9, wherein:the infusion pump further comprises a memory; and the infusion pump is further configured to, prior to delivering the dose of the medicament to the user, store the dosing functions in the memory.

20. A computer-implemented method for operation of an infusion pump, the methodcomprising: obtaining a set of dosing functions at the infusion pump and from a dosing function device, each dosing function representing a dosing target associated with a respective one of activity states, wherein: a user is in one of the activity states at any given time, and the user is capable of transitioning among the activity states during a certain time period; determining state scores at the infusion pump, each of the state scores representing a probability that the user is in a respective one of the activity states; determining, at the infusion pump, a combined dosing function based on the dosing functions and the state scores; andPATENT APPLICATION delivering a dose of medicament to the user via the infusion pump and according to the combined dosing function.

21. The computer-implemented method of Claim 20, wherein:the activity states comprise at least four eating states of the user: not eating, eating start, eating middle, and eating finish; and the state scores for the activity states are computed without any input from the user.

22. The computer-implemented method of Claim 20, wherein the state scores are computedbased on: building a hidden Markov model (HMM) that models the user who changes over time steps as being in one of the activity states at a given time step; determining, for the HMM, a transition matrix including transition probabilities of transitioning between any pair of the activity states; determining preceding state scores at a preceding time step, each of the preceding state scores representing a probability that the user was in a respective one of the activity states at the preceding time step; computing an estimated value based on a measurement on the user; determining emission probabilities at a current time step, each of the emission probabilities is a conditional probability of obtaining the estimated value given that the user is in a respective one of the activity states at the current time step; and computing the state scores at the current time step based on: the transition matrix, the preceding state scores at the preceding time step, and the emission probabilities at the current time step.

23. The computer-implemented method of Claim 22, wherein:the estimated value is computed based on applying a filter on a measurement result on the user from a monitor; and the filter is updated based on a dose of medicament delivered to the user at a previous time step.PATENT APPLICATION24. The computer-implemented method of Claim 20, wherein the combined dosingfunction is generated based on: representing each of the dosing functions based on a respective coefficient set of coefficients corresponding to a respective one of the activity states, wherein each coefficient in the respective coefficient set corresponds to an index; computing weighted coefficients each for a respective index, wherein each of the weighted coefficients is computed based on a weighted average of coefficients corresponding to the respective index in all of the coefficient sets, with corresponding weights being the state scores; and generating the combined dosing function based on the weighted coefficients.

25. A non-transitory, computer-readable medium storing instructions which, whenexecuted by a processor of an electronic device, cause the electronic device to: obtain a set of dosing functions from a dosing function device, each of the dosing functions representing a dosing target associated with a respective one of activity states, wherein: a user is in one of the activity states at any given time, and the user is capable of transitioning among the activity states during a certain time period; determine state scores, each of the state scores representing a probability that the user is in a respective one of the activity states; determine a combined dosing function based on the dosing functions and the state scores; and deliver a dose of medicament to the user according to the combined dosing function.

Images